Proteoglycan UDP-galactose:β-xylose β1,4-galactosyltransferase I is essential for viability in Drosophila melanogaster

Heparan and chondroitin sulfates play essential roles in growth factor signaling during development and share a common linkage tetrasaccharide structure, GlcAβ1,3Galβ1,3Galβ1,4Xylβ1-O-Ser. In the present study, we identified the Drosophilaproteoglycan UDP-galactose:β-xylose β1,4-galactosyltransferase I (dβ4GalTI), and determined its substrate specificity. The enzyme transferred a Gal to the -β-xylose (Xyl) residue, confirming it to be the Drosophila ortholog of human proteoglycan UDP-galactose:β-xylose β1,4-galactosyltransferase I. Then we establishedUAS-dβ4GalTI-IR fly lines containing an inverted repeat of dβ4GalTIligated to the upstream activating sequence (UAS) promoter, a target of GAL4, and observed the F1 generation of the cross between the UAS-dβ4GalTI-IR fly and theAct5C-GAL4 fly. In the F1, double-stranded RNA of dβ4GalTI is expressed ubiquitously under the control of a cytoplasmic actin promoter to induce the silencing of the dβ4GalTI gene. The expression of the target gene was disrupted specifically, and the degree of interference was correlated with phenotype. The lethality among the progeny proved that β4GalTI is essential for viability. This study is the first to use reverse genetics, RNA interference, to study theDrosophila glycosyltransferase systematically.

RNA interference (RNAi) was first recognized in C. elegans as a biological response to exogeneous double-stranded RNA (dsRNA), which induces sequence-specific gene silencing. RNAi is an evolutionarily conserved phenomenon and a multistep process involving the generation of active small interfering RNA (siRNA) in vivo through a reaction with an RNase III endonuclease, Dicer (14 -16). The resulting 21-23-nucleotide siRNA mediates degeneration of the complementary homologous RNA (17). RNAi has recently emerged as a powerful reverse genetics tool to study gene function in many model organisms, including plants, C. elegans and D. melanogaster in which large dsRNAs efficiently induce gene-specific silencing (18,19). Only recently, DNA vector-based siRNA has been reported to suppress the expression of the corresponding gene in mammalian cells (20,21).
In the present study, we identified the Drosophila proteoglycan ␤4GalTI (d␤4GalTI) and performed a biochemical characterization. The protein transferred a Gal to the -␤-Xyl residue, confirming it to be the Drosophila ortholog of human proteoglycan ␤4GalTI (h␤4GalTI). After that, we produced an induc-ible d␤4GalTI RNAi fly using the GAL4-UAS system as a first step toward clarifying the biological role of d␤4GalTI. The d␤4GalTI mRNA was reduced specifically by RNAi, and the severity of the phenotype showed the correlation with the reduction in d␤4GalTI mRNA. The death of the flies proved that d␤4GalTI is essential for viability. This is the first example of the use of reverse genetics to study Drosophila glycosyltransferase systematically.
pVL1393-FLAG-d␤4GalTI and pVL1393-FLAG-h␤4GalTI were cotransfected with BaculoGold viral DNA (Pharmingen, San Diego, CA) into Sf21 insect cells according to the manufacturer's instructions and incubated for 3 days at 27°C to produce recombinant viruses. Sf21 cells were infected with each recombinant virus at a multiplicity of infection of 5 and incubated for 72 h to yield conditioned media containing recombinant ␤4GalTI proteins fused with FLAG peptide. A 5-ml volume of culture medium was mixed with 100 l of anti-FLAG M1 AFFINITY GEL (Sigma). The protein-gel mixture was washed twice with 50 mM Tris-buffered saline (50 mM Tris-HCl, pH 7.4, and 150 mM NaCl) containing 1 mM CaCl 2 and eluted with 100 l of 100 g/ml FLAG peptide in 10 mM Tris-buffered saline (Sigma).
Western Blot Analysis-The enzymes purified above were subjected to 12.5% SDS-polyacrylamide gel electrophoresis, followed by Western blot analysis. The separated proteins were transferred to a Hybond-P membrane (Amersham Biosciences). The membrane was probed with anti-FLAG M2-peroxidase conjugate (Sigma) and stained with Konica Immunostaining HRP-1000 (Konica, Tokyo, Japan). The intensity of positive bands on Western blotting was measured by densitometer to determine the amount of the purified enzyme using FLAG-BAP Control Protein (Sigma).
FIG. 1. cDNA and predicted amino acid sequence of D. melanogaster proteoglycan ␤1,4-galactosyltransferase I. A, the nucleotide sequence and the predicted amino acid sequence. The putative transmembrane domain is underlined. B, the hydrophobicity plot estimated by the method of Kyte and Doolittle with a window of 7 amino acid residues. and Gal␤1,4Xyl␤1-pMph were utilized as acceptor substrates. With 10 nmol of each acceptor, the ␤4GalT activity reaction was performed at both 25 and 37°C for 2 h in 20 l of a reaction mixture containing 14    The asterisks indicate the amino acids identical among all proteins. Conserved amino acids are shown by dots. The three ␤4GalT motifs, including the DXD motif, a metal binding site, are boxed. B, phylogenetic tree of three Drosophila (d␤4GalTI, d␤4GalTA, and d␤4GalTB) and seven human ␤4GalTs (h␤4GalT1 to -6 and h␤4GalTI). d␤4GalTI is the Drosophila ortholog of h␤4GalTI, which is human proteoglycan ␤1,4-galactosyltransferase I, namely human ␤4GalT7. The branch lengths indicate amino acid substitution per site.  ␤4GalTs (d␤4GalTI, d␤4GalTA, and d␤4GalTB) in each d␤4GalTI RNAi fly were determined by competitive RT-PCR. The actual amount of each ␤4GalT mRNA was divided by that of RpL32 mRNA for normalization. The relative amount of each ␤4GalT mRNA to RpL32 mM Hepes buffer (pH 7.4), 0.5% Triton X-100, 11 mM MnCl 2 , 3 M UDP-[ 14 C]Gal (325 mCi/mmol), 250 M UDP-Gal, and 1.15 and 0.44 pmol of purified d␤4GalTI and h␤4GalTI, respectively. The enzyme reaction was terminated by the addition of 400 l of water. After centrifugation of the reaction mixture, the supernatant was applied to a Sep-PakC18 column (Millipore Corp.) equilibrated with water. The unreacted UDP-Gal was washed out with water, and the products were eluted with methanol. The eluates were dried with an N 2 evaporator and dissolved in 30 l of methanol. Then 10 l of the product was applied to an HPTLC plate (Merck) and developed in chloroform/methanol/0.2% CaCl 2 (55:45:10). The bands of reaction products incorporating radioactivity were detected with a BAS2000 Imaging Analyzer system (Fuji, Tokyo, Japan).
RNAi Fly-A cDNA fragment encoding the C-terminal region (nucleotide 685-969 of coding sequence) or the amino-terminal (N-terminal) region (nucleotide 1 to 506 of coding sequence) of d␤4GalTI was amplified by PCR and inserted as an inverted repeat (IR) in a modified Bluescript vector, pSC1, which possesses an IR formation site consisting of paired CpoI and SfiI restriction sites. In all cases, the IR was in a head-to-head orientation. IR-containing fragments were cut out by NotI and subcloned into pUAST, a transformation vector. The cloning procedures will be described elsewhere. 2 The transformation of Drosophila embryos was carried out according to Spradling (22) with w 1118 mutant stock as a host to make 23 UAS-d␤4GalTI-IR fly lines. Then we made one line, N13, which has two copies of the IR, by crossing two lines, N2 and N4, in which the IR was on different chromosomes. Each line was mated with the Act5C-GAL4 fly line, and F 1 progeny was raised at 28°C to observe phenotypes.
Quantitative Analysis of d␤4GalTI mRNA by Competitive RT-PCR-Total RNA was extracted from Act5C-GAL4/UAS-d␤4GalTI-IR and Act5C-GAL4/ϩ third instar larvae and prepupae by the methods of Chomczynski and Sacchi (23). Poly(A) ϩ RNA was isolated from total RNA using Oligotex TM -dT30ϽsuperϾ (Takara Bio Inc.) according to the manufacturer's instructions. First-strand cDNA was synthesized in 50 l of a reaction mixture containing 300 ng of mRNA, 5 mM MgCl 2 , 10 mM dithiothreitol, 0.5 g of oligo(dT) 12-18 , 0.5 mM each dNTP, 40 units of RNasin, and 50 units of Superscript II RT (Invitrogen). After incubation at 50°C for 50 min, the reaction was terminated by heating at 70°C for 15 min, followed by rapid chilling on ice. Competitive RT-PCR of d␤4GalTI was carried out for the region except for the sequences using the IR construction for RNAi. The gene-specific primers used for amplification of d␤4GalTI, d␤4GalTA, d␤4GalTB, and ribosomal protein L32 (RpL32) genes are listed in Table I. The sense and antisense primers for construction of the DNA competitor were prepared by flanking the sequence for amplification of DNA at the 3Ј terminus of each sense and antisense primer for amplification of the target cDNA. The DNA competitors were generated using reagents supplied in the Competitive DNA construction kit (Takara Bio Inc.). Competitive RT-PCRs were performed using 1 l of the first-strand cDNA mixture with a serial dilution of DNA competitor in 50-l reaction mixtures containing a 0.2 M concentration of each of the relevant primers (listed in Table I), 0.2 mM each dNTP, and 2.5 units of TaKaRa Ex Taq. To normalize the efficiency of cDNA preparation among individual samples, measurement of RpL32 mRNA in each cDNA (0.5 l) was carried out using the same competitive RT-PCR method as for d␤4GalTI mRNA. Amplifications involved 30 cycles of 94°C for 30 s, 60°C for 30 s, and 72°C for 30 s. PCR products were subjected to electrophoresis in a 3-4% NuSieve 3:1 agarose gel (Cambrex, Corp., East Rutherford, NJ) and stained with ethidium bromide. ImageMaster VDS-CL (Amersham Biosciences) was used to generate digital images of the agarose gel. The intensities of amplified fragments were quantified using ImageMas-ter analysis software. The amount of target mRNA was estimated from the ratio of the intensity of the competitor band and the target band.

Identification of the Drosophila Proteoglycan ␤1,4-
Galactosyltransferase I-When human h␤4GalTI was used as the query sequence for a TBLASTN search of the Berkeley Drosophila Genome Project, one highly homologous gene, CG11780, was obtained as d␤4GalTI. The complete cDNA of the d␤4GalTI gene and the predicted amino acid sequence are shown in Fig. 1A. The putative protein, consisting of 323 amino acids, was a type II transmembrane protein with a hydrophobic domain in the N-terminal region (Fig. 1B).
When the other members of the human ␤1,4-galactosyltransferase family, h␤4GalT1 to -6, were used as query sequences for the TBLASTN search, two highly homologous genes, CG8536 and CG14517, were also obtained as members of the Drosophila ␤1,4-galactosyltransferase family, d␤4GalTA and d␤4GalTB, respectively. d␤4GalTA and d␤4GalTB showed much lower homology of amino acids to h␤4GalTI (33 and 29%, respectively) than d␤4GalTI to h␤4GalTI (48%).
The ClustalW alignment of the human and Drosophila ␤4GalT families showed that the three ␤4GalT motifs found in the human family, including the DXD motif, a metal binding site, were also conserved in the Drosophila ␤4GalT family ( Fig.  2A) (24,25). A phylogenetic tree of the three Drosophila (d␤4GalTI, d␤4GalTA, and d␤4GalTB) and seven human ␤4GalTs (h␤4GalT1 to -6 and h␤4GalTI) was generated based on the amino acid sequences (Fig. 2B). d␤4GalTI was confirmed to be the Drosophila ortholog of h␤4GalTI. Both d␤4GalTA and d4␤GalTB showed higher homology to h␤4GalT1 to -6 than h␤4GalTI.
Characterization of the Galactosyltransferase Activity of d␤4GalTI-The FLAG-tagged recombinant d␤4GalTI was expressed in insect cells to determine whether or not d␤4GalTI has galactosyltransferase activity. The soluble form was prepared by replacing the N-terminal region including the cytoplasmic and transmembrane domains, amino acids 1-35, with an Ig signal sequence and FLAG peptide sequence. The secreted enzyme was purified with anti-FLAG M1 gel and quantitated by Western blotting analysis using Anti-FLAG antibody. FLAG-tagged recombinant h␤4GalTI was also prepared by the same procedure.
The purified enzymes were used for a galactosyltransferase assay with various acceptor substrates (Table II). The determined amounts of d␤4GalTI and h␤4GalTI and the same amount of each substrate were used for the enzyme reactions, so we could determine relative activities that were comparable. h␤4GalTI had strong activity toward the -␤-Xyl residue, whereas it had only slight activity toward -␣-Xyl and no activity toward -␤-GlcNAc, -␤-Glc, -␤-Gal, and -␤-GalNAc as reported previously (6,7). d␤4GalTI also showed strong activity toward the -␤-Xyl residue; only slight activity toward -␣-Xyl; and no activity toward -␤-GlcNAc, -␤-Glc, -␤-Gal, and -␤-Gal-NAc. These results demonstrated that d␤4GalTI was the Dro-2 R. Ueda and K. Saigo, manuscript in preparation. mRNA in F 1 progeny of the w 1118 crossed with Act5C-GAL4 fly, Act5C-GAL4/ϩ, which corresponds to the wild type, was presented as 1. Amounts of 1 and 0.5 l of synthesized cDNA were used for the quantitation of ␤4GalTs and RpL32, respectively. The cDNAs for each ␤4GalT mRNA and RpL32 mRNA were amplified together with 0.5-100 ϫ 10 5 copies and 2.5-10 ϫ 10 7 copies, respectively, of the corresponding competitor DNAs. A, the mRNA levels of three kinds of ␤4GalTs in the third instar larvae of the F 1 progeny of each N line of UAS-d␤4GalTI-IR fly crossed with Act5C-GAL4, designated as Act5C-GAL4/N. Each N line has the IR of the sequence encoding the N-terminal region of d␤4GalTI. Act5C-GAL4/N2, Act5C-GAL4/N4, and Act5C-GAL4/N13 showed lethality at the pupal stages, whereas Act5C-GAL4/N6 was viable and morphologically normal. The N13 line has two copies of the IR on chromosomes 2 and 3. B, the mRNA levels of three kinds of ␤4GalTs in the third instar larvae of the F 1 progeny of each C line of UAS-d␤4GalTI-IR fly crossed with Act5C-GAL4 fly, designated as Act5C-GAL4/C. Each C line has the IR of the sequence that encodes the C-terminal region of d␤4GalTI. Act5C-GAL4/C5 and Act5C-GAL4/C6 were pupal lethal. C, the mRNA level of d␤4GalTI in the prepupae of Act5C-GAL4/N4 and Act5C-GAL4/N13. sophila ortholog of h␤4GalTI in view of their activities. But the ␤4GalT activity of d␤4GalTI toward the -␤-Xyl residue was almost half that of h␤4GalTI at both 25 and 37°C.
Viability of Inducible d␤GalTI RNAi Flies-Proteoglycan ␤4GalTI contributes to the synthesis of the common carbohydrate-protein linkage structure, GlcA␤1,3Gal␤1,3Gal␤1,4Xyl ␤1-O-Ser, of proteoglycan including heparin/HS and CS/DS. If proteoglycan ␤4GalTI is inactivated, every proteoglycan lacks GAG, and a severe biological phenotype is expected. To test this hypothesis, we tried to make inducible d␤4GalTI RNAi flies according to the method described under "Experimental Procedures." A scheme of the heritable and inducible RNAi system is shown in Fig. 3. In this report, we used Act5C-GAL4 as a GAL4 driver to induce d␤4GalTI gene silencing in all cells of the fly. The Act5C-GAL4 fly has a transgene containing yeast transcriptional factor GAL4, the expression of which is under the control of the cytoplasmic actin promoter. 24 UAS-d␤4GalTI-IR fly stocks having a transgene containing two types of the IR of d␤4GalTI ligated to the UAS promoter, a target of GAL4, were established (Table III). The IR of d␤4GalTI was separated by an unrelated DNA sequence that acts as a spacer to give a hairpin loop-shaped RNA. C-1 to C-11 have a transgene containing the IR of the sequence encoding the C-terminal region of the catalytic domain (amino acids 209 -322). N-1 to N-13 have a transgene containing the IR of the sequence encoding the N-terminal region (amino acids 1-167). In the F 1 generations of the Act5C-GAL4 fly and the UAS-d␤4GalTI-IR fly, dsRNA of d␤4GalTI is expressed ubiquitously under the control of the cytoplasmic actin promoter to induce d␤4GalTI gene silencing.
The phenotypes of the F 1 of each UAS-d␤4GalTI-IR fly crossed with the Act5C-GAL4 fly are shown in Table III; 65% (15 of 23) of these crosses caused lethality in the progeny (i.e. the flies could not develop into adults). The expression of d␤4GalTI dsRNA by crossing the N13 line carrying two copies of UAS-d␤4GalTI-IR to the Act5C-GAL4 fly also led to lethality at the pupal stages. These results clearly demonstrated that proteoglycan ␤4GalTI is essential for the viability of flies.
Quantitative Analysis of d␤4GalT mRNA in Each Inducible d␤4GalTI RNAi Fly by Competitive RT-PCR-To test the efficiency and specificity of RNAi in this system, the mRNA levels of all Drosophila ␤4GalTs (d␤4GalTI, d␤4GalTA, and d␤4GalTB) were determined in each d␤4GalTI RNAi fly by competitive RT-PCR (Fig. 4). The relative amount of each ␤4GalT mRNA to RpL32 mRNA in F 1 progeny of w 1118 crossed with the Act5C-GAL4 fly, Actin5C-GAL4/ϩ, which corresponds to the wild type, was presented as 1. The F 1 progeny of each N or C line of the UAS-d␤4GalTI-IR fly crossed with Act5C-GAL4 fly was designated as Act5C-GAL4/N or Act5C-GAL4/C, respectively.
N lines have a transgene including the IR of the sequence encoding the N-terminal region of d␤4GalTI. N13 having two copies of the IR on chromosomes 2 and 3 was made from the N2 and N4 lines. The degree of expression of the transgene is known to depend on its sites of insertion on the chromosome. Act5C-GAL4/N2, Act5C-GAL4/N4, and Act5C-GAL4/N13 were pupal lethal, whereas Act5C-GAL4/N6 was viable and morphologically normal (Table III). First, we determined the amounts of the three kinds of d␤4GalTs mRNA in the third instar larvae of these four RNAi flies and the wild-type fly, Act5C-GAL4/ϩ. The ratios of reduction in d␤4GalTI mRNA of Act5C-GAL4/N13, Act5C-GAL4/N4, Act5C-GAL4/N2, and Act5C-GAL4/N6 were 0.26, 0.32, 0.36, and 0.76, respectively, demonstrating a correlation with the severity of the pheno-type (Fig. 4A). F 1 progeny of the N13 line having two copies of the IR had less d␤4GalTI mRNA than those of the N2 line and N4 line, which were crossed to make the N13 line. Reductions in d␤4GalTA mRNA and d␤4GalTB mRNA were not observed in all RNAi flies. It was clearly demonstrated that the d␤4GalTI mRNA was disrupted specifically, and the ratio of degraded d␤4GalTI mRNA was well correlated with the severity of the phenotype.
Similar analyses were performed for the F 1 progeny of C lines crossed with Act5C-GAL4 fly, Act5C-GAL4/C5, and Act5C-GAL4/C6. Both RNAi flies showed lethality at the pupal stages (Table III). The d␤4GalTI mRNAs in the third instar larvae were also interfered with specifically, and the ratios of degraded d␤4GalTI mRNA of Act5C-GAL4/C5 and Act5C-GAL4/C6 (0.35 and 0.45, respectively) were almost the same as those of Act5C-GAL4/N2 and Act5C-GAL4/N4 (Fig. 4B). The efficiency of RNAi did not depend greatly on the target sequences using the constructions of IR.
We also determined the amount of d␤4GalTI mRNAs in prepupae of Act5C-GAL4/N4 and Act5C-GAL4/N13 (Fig. 4C). The target efficiency in the prepupae was almost the same as that in the third instar larvae.
The above results clearly demonstrated that the expression of the target gene was specifically reduced by RNAi in this Drosophila RNAi system to induce the phenotype. DISCUSSION We identified the Drosophila proteoglycan ␤4GalTI by molecular and biochemical analyses and then made the RNAi fly to investigate d␤4GalTI function in vivo. The expression of the target gene was disrupted specifically in the RNAi fly and the degree of interference was correlated to phenotype. The reduction of d␤4GalTI mRNA caused lethality, indicating an essential function of d␤4GalTI for viability. This is the first example of a reverse genetics approach to the systematic study of Drosophila glycosyltransferase.
Drosophila has three members of the ␤4GalT family, d␤4GalTI, d␤4GalTA, and d␤4GalTB ( Fig. 2A). The phylogenetic tree (Fig. 2B) and acceptor substrate specificities (Table  II) of the ␤4GalT family clearly demonstrated that d␤4GalTI is the Drosophila ortholog of h␤4GalTI. d␤4GalTA and d␤4GalTB showed higher homology to h␤4GalT1 to -6 than to h␤4GalTI, but the Drosophila ortholog of the six h␤4GalTs could not be identified. So there is a possibility that d␤4GalTA and d␤4GalTB share acceptor substrates, to which h␤4GalT1 to -6 can transfer Gal. Recently, h␤4GalT1 has been reported to transfer Gal to fringe-modified O-fucose glycans on the Notch protein and the elongation of glycans was necessary to modulate Notch signaling. A novel Drosophila galectin has also been isolated (26). It is still unknown which of the two d␤4GalTs works to elongate O-fucose glycans on Notch or synthesize the ligands of Drosophila galectin. We are now attempting to determine the substrate specificity of the two d␤4GalTs.
The three members of the d␤4GalT family also conserved the three ␤4GalT motifs found in the h␤4GalT family ( Fig. 2A) (25). The crystal structure of the bovine ␤4GalT1 has already been reported (24,27,28). The DXD motif is a Mn 2ϩ -binding site, and the other two motifs expose the surface of the catalytic pocket. The FNRA motif is involved in UDP-Gal binding and the negatively charged residues of the GWGXEDD(D/E) motif contribute to UDP-Gal and UDP-glucose binding. The three motifs conserved between human and Drosophila had amino acid sequences that related to the binding of metal or donor substrates.
d␤4GalTI showed roughly half the activity of h␤4GalTI toward each substrate at both 25 and 37°C, although the sub-strate specificities of the two were similar (Table II). The breeding temperature of flies and the body temperature of humans are 25 and 37°C, respectively. Under physiological conditions, h␤4GalTI performed a more efficient reaction than d␤4GalTI. Very recently, two papers about d␤4GalTI have been published (29,30) reporting similar enzymatic activity to ours.
We made 24 UAS-d␤4GalTI-IR fly lines and observed the F 1 generation of each UAS-d␤4GalTI-IR fly crossed with the Act5C-GAL4 fly. The severity of the phenotype differed between the stocks. Approximately 65% of the flies died at the pupal stages (Table III), but some lived to become adults, similar in morphology to the wild type. Because the degree of expression of the transgene is known to depend on its sites of insertion on the chromosome, it is reasonable that the phenotypes differed. The reduction in d␤4GalTI mRNA was correlated with the severity of the phenotype (Fig. 4A). The severest phenotype should be considered to represent the real phenotype of the mutant. Although we have no data indicating that cell, tissue, or organ abnormalities caused the death of individual flies, finer analyses of the phenotype should reveal the in vivo function of d␤4GalTI.
We analyzed the amounts of three d␤4GalT mRNAs (d␤4GalTI, d␤4GalTA, and d␤4GalTB) to estimate the specificity and efficiency of RNAi in our Drosophila-inducible RNAi system. The RNAi occurred only on d␤4GalTI and had no effect on the other members of the d␤4GalT family, d␤4GalTA and d␤4GalTB (Fig. 4, A and B). During the process of RNAi, 21-23-nucleotide siRNA mediates the degeneration of the complementary homologous RNA (17). If even one nucleotide differs between the siRNA and target mRNA, siRNA cannot mediate the degeneration of target RNA (17). Comparing the DNA sequence of d␤4GalTI with the sequences of d␤4GalTA and d␤4GalTB, we could not find any identical regions longer than 21 nucleotides. This must be the reason why the RNAi of d␤4GalTI was specific with no cross-effect for d␤4GalTA and d␤4GalTB.
The efficiency of RNAi largely did not depend on the target sequences using the constructions of IR (Fig. 4, A and B), and the ratio of degraded d␤4GalTI mRNA was well correlated with the severity of the phenotypes. These findings demonstrate that our Drosophila inducible RNAi system has the potential to become a powerful tool for analyses of the biological roles of glycosyltransferase. However, one might have to make an effort to increase the efficiency of RNAi. A small amount of maternal RNA might remain even after the occurrence of RNAi.
d␤4GalTI contributes to the synthesis of the common linkage structure of heparin/HS and CS/DS. When d␤4GalTI is inactivated in the RNAi fly, levels of both GAGs are reduced. Our results clearly demonstrated that GAGs on core proteins are important to viability (Table III). It has been reported that the h␤4GalTI gene of patients with Ehlers-Danlos syndrome has mutations that reduce ␤4GalTI activity (7,31). The patients show an aged appearance, developmental delay, dwarfism, craniofacial disproportion, generalized osteopenia, and various connective abnormalities. The RNAi fly of d␤4GalTI should serve as a model of this kind of disease.
D. melanogaster is well established as a model for genetic analysis. Recently, fly mutants with severe phenotypes related to developmental processes have begun to be used to address the defects of glycosyltransferases. As one example, the Drosophila fringe gene has been demonstrated to encode an Ofucose ␤1,3-N-acetylglucosaminyltransferase that extends the O-fucose moieties on Notch to modulate Notch activation by the ligands, Delta and Serrate/Jagged (32,33). Some flies with mutations related to proteoglycan, dally (34), sugarless, tout-velu, and sulfateless (2), have demonstrated defects in signaling of the growth factors including Wingless, Decapentaplegic, Hedgehog, and fibroblast growth factors. Recently, one recessive lethal mutant has been reported to have a missense mutation causing a reduction of UDP-GalNAc:polypeptide N-acetylgalactosaminyltransferase activity (35,36). Considering the recent progress made in studies of Drosophila glycans as mentioned above, D. melanogaster will become a powerful tool for analysis of the biological roles of glycans.
In this report, we demonstrated the first systematic reverse genetics approach using RNAi of Drosophila glycosyltransferase and showed that RNAi worked well in the case of a glycosyltransferase. We found almost 70 Drosophila glycosyltransferases by performing a TBLASTN search of the Drosophila data bases using mammalian glycosyltransferases as the query sequence. It is possible to make a RNAi fly for each of the 70 glycosyltransferases, whereas knock-out mice cannot be made. The inducible glycosyltransferase RNAi fly obtained using the GAL4-UAS system will open the way for the analysis of the biological role of glycans.